Print PageThe migratory movement of animals represents one of the natural worlds amazing spectacles. The study of such movement has recently gained significant momentum. As technological innovation has reduced the size and cost of various telemetry and biologging devices, the number of species and individuals being tracked has grown significantly. This growth is necessarily accompanied by an explosion in the volume of data gathered, and often in the richness of tracking information available. Such data can be hard to interpret using traditional analysis, but lends itself well to the application of machine learning and pattern recognition techniques.
Results/Conclusions
We present a variety of approaches to the investigation of at-sea behaviour in migratory seabirds using machine learning. We examine the objective identification of behavioural states during foraging behaviour, and explore the combination of unsupervised and supervised classification to identify marine stopovers during migratory journeys. We then present recent work on combining these approaches in multiple data series to gain a more detailed understanding of wintering distributions, migration and foraging behaviour. We discuss how these methods can be used to approach large datasets, and can reveal novel understanding of underlying behaviour, and finally how they may also allow researchers to reduce the negative impacts of larger devices.
